Pleasant@Rutgers.arpa (01/24/83)
HUMAN-NETS Digest Tuesday, 25 Jan 1983 Volume 6 : Issue 6
Today's Topics:
Technology - Combinations of Telephones and Terminals &
WorldNet (5 msgs),
Computers and People - Human Memory Capacity (2 msgs) &
New Interactive Job Search Service
----------------------------------------------------------------------
From: "REX::MINOW c/o" <DEC-HNT at DEC-MARLBORO>
Date: 27 Dec 1982 1930-EST
Subject: Touch-Tone to Ascii Conversion
Several recent Human Nets messages have discussed generation of
Ascii by means of a DTMF (Touch-Tone) keypad. I've done a bit of
work on this and hope the following might be of interest:
First, holding multiple buttons down at the same time probably won't
work in the real world. There are several companies offering DTMF
decoders (coupled with FCC approved telephone line interfaces) which
are generally set to reject single frequencies as required by the
AT&T specifications.
Using multiple keystrokes seems to offer the best of a bad situation
(clumsy, but workable). Several such systems have been done. For
example, there is a very nice automated weather forecast system
using synthesized speech and DTMF control done by the FAA. Also,
Lauren Weinstein implemented a telephone interface to Unix at UCLA,
using the Unix speak program (text to speech for a Votrax ML1) and a
Bell 407 telephone line interface. With much help from Lauren, I
implemented a telephone interface to RSTS/E about 3 years ago using
the NRL text- to-speech system. All three systems used essentially
the same DTMF to Ascii encoding method:
Letters are entered by pressing the button containing the letter,
followed by a button indicating which of the three (left, middle, or
right) letters is desired. Thus ABC would be 21, 22, 23. The FAA
system accepted only 1/2/3 for the second button, while the other
systems allowed "any number in that column". Thus, on the UCLA and
DEC systems, "HUMAN" could be encoded 45, 88, 64, 21, 65.
There are two letters missing from the keypad. The DEC system put
them on the '1' key as "<space>QZ" (The other systems used
something similar, but I felt that 11 was a good way to encode
space.)
Digits were encoded in the DEC system by combining them with the
ZERO key. Since I could never remember whether the zero came first
or last, my program accepted either encoding.
Now, the fun begins... The SHARP key was used for control characters:
#1 Z == end of file (CTRL/Z at Dec), #2 C == CTRL/C,
#3 D == Delete (rubout), #6 O == CTRL/O (Cancel output)
#7 R == Retype line (CTRL/R) #7 U == CTRL/U (Delete line)
## == Carriage return.
The STAR key was used for control functions. Lauren and I
implemented case shifts and locks and numeric, control, and 8-bit
octal input. There was also a punctuation mode (courtesy of Lauren)
whereby the next three button pushes were interpreted as a graphic
character. For example, 365 (DOL) for '$', 758 (PLU) for '+', 277
(BSP) for backspace, etc. Many characters had several definitions.
For example '<' was both 522 (LAN) and 535 (LES). Finally, there
were a few predefined messages:
910 Logout
911 MAIL
990 run games:dungeon
While it was a nice toy and a fun demo, and once in a while was very
useful, the amount of button pushing you had to go through was
extremely frustrating. Also, the quality of the Votrax voice was
not satisfactory for anything more than games playing.
I'd appreciate hearing with anyone with ideas on improving this
system; especially someone who would have no other access to a
computer.
Finally, the IBM voice mail system uses the keypad to enter user
names. They use the digits (MINOW would be entered 64669) as a hash
function. On the IBM system, Q is on the 7 key (PQRS) and Z on the
9 key (WXYZ).
Martin Minow
------------------------------
Date: 20 Jan 83 16:42:43 EST (Thu)
From: Chris Torek <chris.umcp-cs@UDel-Relay>
Subject: Worldnet & Lawyers
I feel that if there is to be a legal stand on whether computer mail
is to be considered the "written word" or the "spoken word", we're
going to have to call it the latter. I certainly wouldn't write
down a lot of the things that I type in. (But then, I don't like
writing anyway. I type everything -- it's easier [at least with a
computer!].)
------------------------------
Date: 21 January 1983 04:08 est
From: SSteinberg.SoftArts at MIT-MULTICS
Subject: Changing face of u-Computing.
I have to agree with POURNE at MIT-MC, it is NOT "too late for
parents" and one of the biggest problems is that a lot of stuff is
not explained in English. Arguing that most of us are not
automobile mechanics is not valid, using a car effectively does NOT
involve knowing how to take it apart, but rather how to make it do
exactly what you want. Anyone who tells me I cannot direct dial to
London because I don't even know if my call is sent via cable or
satellite is incorrect, because I can, just as I can use calculus
with almost NO knowledge of analysis.
As far as language goes I keep finding that I can rephrase things in
English and make myself understood, even when explaining rather
complex issues (e.g. data compression, garbage collection,
inheritance of procedural information). The description often does
sound "funny" and possibly stilted and reminds me of the plain
language automobile insurance policy my insurance company sent me.
Not only did they leave out all the "pursuants" and give the
sections names instead of numbers but they actually made it clear
that they would pay for my new windshield and not stick me for the
$250 deductible. Let's face it, I can tell a carpenter that my door
is off the hinges and let him worry about the gudgeon and the
pintel.
------------------------------
Date: 21 January 1983 04:47 EST
From: Jerry E. Pournelle <POURNE @ MIT-MC>
Subject: The changing face of Micro-computing...
Anything not worth doing is not worth doing well.
Anything worth doing well is worth doing for money
------------------------------
Date: 23 January 1983 03:57 EST
From: Robert Elton Maas <REM @ MIT-MC>
Subject: changing face of Micro-computing / have to learn too much
Subject: first
It seems to me if information databases were properly structured and
had adequate access methods, one wouldn't have to "learn a lot more
than really necessary" to get some task done. It should be possible
to ask a question "how do I do ...?" and get back a precise answer.
If any terminology is unknown or if the overall frame is unfamiliar,
it should be possible to ask for definitions of the terminology or
for general background information or to ask specific questions
about the frame. It shouldn't be necessary to first learn immense
crud and only then get an answer to the original question.
The world-net should include an information system that includes
everything precisely known by anyone who either published it
traditionally or added it directly to the information database, and
which permits top-down learning, ask the question and get a direct
answer and fill in any gaps immediately later rather than first
learn all the gaps and last the precise answer.
Has anybody experimented with a top-down question-answering
database, on the net or elsewhere? I'd like to see the idea tried.
------------------------------
Date: 23 January 1983 04:05 EST
From: Robert Elton Maas <REM @ MIT-MC>
Subject: The changing face of Micro-computing / jargon vs. english
Regarding the dilemma, if you use jargon nobody can understand it,
if you don't the answer comes out muddled. The solution is:
(See my top-down answer-to-question message I sent to Pournelle and
humnet.) Answer with jargon when necessary, but plenty of precise
technical language. True hack jargon like half the junk in the AI
Jargon file (glork, jiffie, state-of-the-worldP, etc.) shouldn't be
used, but nice precise words like "algorithm" and "process" and
"pipeline" certainly should be used where appropriate rather than
trying to find common English words to replace them. Then provide
definitions for any terminology (jargon or technicalese) the student
doesn't already know and can't guess at from context. -- It'll take
some training for a student to get used to filling in the blanks by
his/her action of asking a lot of questions, rather than just
accepting the definition as presented without understanding it, but
ultimately this teaching method should work fine.
------------------------------
Date: 20 Jan 1983 09:35:16-EST
From: John McLean <mclean@NRL-CSS>
Subject: infinite memory
Assuming that memory works by changing a (finite number of) discrete
parts (neurons, atoms, or whatever) of a human in one of a finite
number of ways, then memory must be finite. Further, I'm not sure
how else memory could work. However, neither am I sure what a claim
of finite memory amounts to.
My confusion stems from the fact that I can *in theory* recognize an
infinite number of integers. (Of course the above argument
establishes that, in fact, there are numbers large enough that even
if I could live long enough to compare them digit by digit, I could
not tell them apart). It may be objected that I only memorized a
finite number of rules that enabled me to "memorize" an infinite
number of integers, but that's what conceptualization is all about.
The point might be more convincing if I learned the integers by
being shown all of them and memorized them all by subconsciously
abstracting the rules for myself. But how could I ever be shown an
infinite number of integers in a finite time, given that I need a
discrete amount of time for each one to register? More generally,
how can anyone ever be exposed to an infinite amount of discrete
information that must be memorized?
Since I assume that psychologists must have the concepts of memory,
information, etc. more pinned down than I do, I would appreciate any
response that would clear up my confusion.
Thanks,
john
------------------------------
Date: 20 Jan 1983 0935-PST
From: LAWS at SRI-AI
Subject: Dot Pattern Memory
I have read about a man who has memorized the entire night sky as
visible through powerful binoculars. He scans the sky for hours
nearly every night, and has discovered several comets and other
events. (He regrets one near miss where a patch of the sky just
didn't look right, but he was unable to identify the interloper. A
Japanese astronomer got credit for the discovery.)
I am sure that the variable brightnesses of the stars, and
particularly the visible/normally invisible distinction, were a
great help in learning the star patterns.
-- Ken Laws
------------------------------
Date: 22 Jan 1983 1533-EST
From: "FLOPPY::MRBILL c/o" <DEC-HNT at DEC-MARLBORO>
Subject: New Interactive Job Search Service
Of course, it was only a matter of time...
A new radio add has been blitzing the Boston area this week,
advertising a new job search service. The adds sparked my
curiosity, and I called up the service for further information.
This new service is a dial-up computer. A job seeker simply logs
into a "large mainframe computer," and starts searching the
"electronic data base" for a dream job. They claim to have listings
from companies ranging from the Fortune 500 to new startups.
Strict confidentiality is guaranteed. To register for the service,
simply call up the service, and they will provide you with an
account number (not name!) and password. From that point on it's
you, your terminal, and their computer. (And $15 for every 2 hours
of connect time.)
When you login, the service is menu driven. You can enter the type
of job that you are looking for. Presumably, it is a "check of the
box(s) that most closely match what you are looking for" type of
thing, but with a menu structure.
Any job descriptions matching your interests will be displayed. If
you have any additional questions on any of the listings, you can
send electronic mail directly to the company that submitted the
listing, and receive a reply "within minutes(!???)" If you are
interested in the job, you can release your resume to the company.
(There are on-line resume creation aids.) Your resume is strictly
your own. It will not be released to any company unless you request
it.
Computers have often been used by job services and headhunters, but
this is the first that I have heard of job seekers allowed hands on
interactive computing. Has anyone out there ever used a service
like this?
-mr. bill
------------------------------
End of HUMAN-NETS Digest
************************